Power grid load management method and system based on big data

Abstract

The invention discloses a power grid load management method and system based on big data, and relates to the field of power grid management, and the method comprises the following operation steps: S1, multi-data collection and deep integration; s2, data analysis and intelligent modeling; s3, distributed energy virtual power plant cooperative operation; and S4, establishing a demand response incentive mechanism based on the block chain. According to the big data-based power grid load management method and system, social media public opinions and power utilization are associated and mined, the social media platform and the local life forum are dynamically monitored in real time, sudden change of power utilization load caused by urban activities and livelihood hotspots is quickly captured, a mobile energy storage vehicle can be scheduled in advance to guarantee power supply, and the power utilization efficiency is improved. According to the method and the system, the energy storage can be adjusted through an intelligent regulation and control plan, the electric power can be distributed according to the actual load condition of the power grid, and the electric power supply can be effectively ensured.

Description

Technical Field

[0001] The present invention relates to the field of power grid management, and particularly to a power grid load management method and system based on big data. Background Art

[0002] The power grid is a complex power transmission and distribution system, including power generation facilities, transmission lines, and substations. In power consumption areas such as cities, towns, and industrial parks, the power grid distributes electric energy from substations to each block, factory, and residential user. The distribution lines are like the veins of a city, delivering electric energy to every corner where it is needed.

[0003] When conducting load management of the power grid, most are limited to internal data of the power system, such as electricity meter readings and substation operation data, and cannot comprehensively understand the electricity consumption demand. As a result, when conducting load management, accurate load management cannot be carried out according to the usage demand, and it is difficult to cope with complex and changeable electricity consumption scenarios, and cannot adapt to load changes under different seasons, weather conditions, and social activities.

[0004] Therefore, it is very necessary to propose a power grid load management method and system based on big data to solve the above problems. Summary of the Invention

[0005] The main object of the present invention is to provide a power grid load management method and system based on big data, which can effectively solve the problems in the background art.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A power grid load management method based on big data includes the following operation steps:

[0008] S1: Multi-data collection and in-depth integration, collecting and integrating cross-industry data and social media data, and planning the power grid based on the collected data to improve the working load of the power grid;

[0009] S2: Data analysis and intelligent modeling, collecting historical electricity consumption data, and constructing models and knowledge graphs based on the collected data to predict future power grid electricity consumption load conditions, and at the same time optimizing the operation and service strategies of the power grid;

[0010] S3: Cooperative operation of distributed energy virtual power plants, establishing an energy database and a data collaborative scheduling platform, integrating external data and the energy database in real time, and then adjusting the power generation plan of energy in advance, and storing and scheduling the power grid electricity according to the actual situation;

[0011] S4: Establish a blockchain-based demand response incentive mechanism. Utilize the distributed ledger feature of the blockchain to record all behavior data used for participating in demand response, and design an incentive points system to encourage users to actively reduce electricity consumption during peak hours.

[0012] Preferably, in S1, the cross-industry data includes cooperating with financial institutions, actively communicating and sharing data, regularly obtaining the credit limit and capital flow of enterprises from financial institutions. Based on the obtained data, through data analysts, according to industry experience and past data models, judge the dynamic production scale of enterprises. When the production scale expands, it can provide accurate guidance for the layout optimization of the power grid.

[0013] The social media data includes using natural language processing technology tools to access the public data interfaces of social media platforms and local life forums, setting keywords including city names and activities, home appliance brands and promotions, and real-time capturing relevant discussion content. Once it is monitored that a large-scale activity is to be held in a certain area, combine the historical electricity consumption data of similar activities in the area and the activity heat evaluation algorithm to calculate the newly added electricity load around the activity, and based on the calculated electricity load, give an electricity consumption reminder for the current area, immediately notify the power grid dispatching department to arrange mobile energy storage vehicles, and at the same time send off-peak electricity consumption reminders to surrounding residents through text messages and APP push. The activity heat evaluation algorithm includes the calculation of the social media influence score S, the calculation of the ticket heat score T, and the calculation of the local life service heat score L. Among them, the calculation formula for the social media influence score S is:

[0014]

[0015] where α is the social media weight coefficient, set according to experience and historical data regression analysis, indicating the relative importance of social media in the overall heat evaluation; Like is the number of likes; Comment is the number of comments; Retweet is the number of retweets; w Like is the weight of likes; w comment is the weight of comments; w Retweet is the weight of retweets; i is the number of posts about the activity on social media;

[0016] The calculation formula for the ticket heat score T is:

[0017] T = β(TicketSold × w TicketSold - TicketRefund × w TicketRefund ) + γTicketSpeed × w TicketSpeed ;

[0018] where β is the ticket sales volume weight coefficient; TicketSold is the ticket sales volume; TicketRefund is the ticket refund volume; w TicketSold is the weight of the ticket sales volume; wTicketRefund is the weight of the refund volume; γ is the weight coefficient of the sales speed; TicketSpeed is the sales speed; w TicketSpeed is the weight of the sales speed;

[0019] The calculation formula for the local life service popularity score L is:

[0020] L = δ

[0021] (Collect × w collect + Appointment × w Appointment + Browse × w Browse );

[0022] where δ is the local life service platform weight coefficient; Collect is collection; Appointment is reservation; Browse is browsing; w collect is the weight of collection; w Appointment is the weight of reservation; w Browse is the weight of traffic;

[0023] The total activity popularity H is calculated based on the social media influence score S, the ticket popularity score T, and the local life service popularity score L. The formula is:

[0024] H = S + T + L;

[0025] The higher the calculated H value, the greater the activity popularity. Furthermore, the possible additional electricity load around the activity can be predicted based on the popularity, and the power grid dispatching preparation can be made in advance;

[0026] Continuously pay attention to the hotspots in the electricity consumption fields such as new energy vehicles and energy-saving household appliances, analyze the changes in the public discussion popularity, and if it is found that a certain electricity-consuming device is popular, quickly cooperate with the manufacturer to implant the intelligent electricity consumption mode program in the electricity-consuming device sales link to ensure that after purchase, the electricity-consuming device can automatically pre-cool according to the low load period of the power grid, and at the same time match the user with a preferential electricity price package to guide the user to use electricity during off-peak hours.

[0027] Preferably, in S2, it specifically includes generating adversarial network-assisted load forecasting and knowledge graph-driven user electricity consumption portrait construction. Among them, the generating adversarial network-assisted load forecasting includes the following operation steps:

[0028] S201: Data preparation, collect the historical electricity consumption data of the power grid, covering the electricity load information under different seasons, weekdays and holidays, and various weather conditions, and at the same time sort out the factors affecting the electricity load. Mark the above data accurately according to the time series, and divide the collected data into training set, validation set and test set according to the ratio of 70%, 15%, and 15% to ensure uniform data distribution and comprehensive reflection of various electricity consumption scenarios;

[0029] S202: Model construction. Build a generative adversarial network architecture, including a generator and a discriminator. The generator part is constructed based on a deep learning neural network and is used to attempt to generate data simulation samples similar to real electricity consumption scenarios. The samples cover the key features of electricity load magnitude and change trends. The discriminator is constructed based on a neural network. The inputs are the simulation samples generated by the generator and real historical electricity consumption data samples. By extracting and comparing features, it outputs a discrimination probability, which is used to represent the likelihood that the input sample is a real data sample or a simulation data sample.

[0030] S203: Training and optimization. Let the generator and the discriminator conduct adversarial training. During the training process, use the validation set to regularly evaluate the model performance and monitor the similarity index between the simulation data samples and the real data samples. When the performance of the model on the validation set no longer improves, stop the training. Use the test set to conduct the final evaluation of the model. Compare the predicted electricity load of the model with the actual electricity load, and calculate the accuracy rate and recall rate. When the expected effect is achieved, deploy the model to the actual power grid load prediction system to enable the power grid to perform load prediction under abnormal conditions.

[0031] The construction of a user electricity consumption portrait driven by a knowledge graph includes the following operation steps:

[0032] S204: Knowledge graph data collection. Collect the basic electricity consumption information of users from the internal system of the power company, including user ID, address, monthly electricity consumption, and payment records. At the same time, obtain the detailed electricity consumption data recorded by the user's electric meter, including the peak and valley periods of daily electricity consumption and power factor. At the same time, cooperate with the public security department to obtain user age and family member information, cooperate with the real estate department to obtain the property area and housing type structure, and cooperate with financial institutions to understand the user's credit status to ensure that the data source is legal and compliant. Collect the information of the electricity-consuming equipment owned by the user. On the one hand, through the registration records when the user purchases electrical appliances, and on the other hand, use smart meters to identify the characteristics of newly added electricity-consuming equipment.

[0033] S205: Knowledge graph construction. Determine the entities and relationships of the knowledge graph. The entities include users, electricity-consuming equipment, electricity consumption areas, and policies and regulations. The relationships include that a user owns electricity-consuming equipment, a user is located in a certain electricity consumption area, and a user applies a certain policy. Based on graph database technology, store and organize the collected data according to the determined entities and relationships to form an interconnected knowledge graph.

[0034] S206: Generation of user electricity consumption profiles. Based on the knowledge graph, extract the key features of users, including age group, credit rating, and usage of new energy equipment. Classify users according to different feature combinations, and formulate personalized electricity consumption service strategies for different categories of users. The strategies include increasing the peak-valley electricity price discounts for elderly users with high credit and multiple new energy devices at home, and guiding them to use equipment with flexible electricity consumption time adjustment during the low valley electricity price period at night;

[0035] For young startup enterprises, design flexible electricity settlement methods based on the industry prospects and financial status associated with the knowledge graph, including allowing enterprises to settle electricity bills quarterly and providing a certain amount of electricity bill deferral period, and then adjusting to a more conventional settlement mode later;

[0036] Provide an intelligent equipment management solution for industrial enterprises, and automatically adjust the start-stop sequence and operation time of large equipment according to the production process and grid load conditions.

[0037] Preferably, in S3, the following operation steps are specifically included:

[0038] S301: Conduct a comprehensive inventory of all distributed energy resources within the grid area, and record in detail the installed capacity, power generation efficiency, equipment health status, historical power generation data, and grid connection nodes of each energy site to establish a detailed distributed energy database;

[0039] S302: Develop a specialized big data collaborative scheduling platform, and use the Kalman filter algorithm to integrate external data such as weather forecasts, hydrological monitoring, and biomass raw material supply with the information in the energy database in real time;

[0040] S303: When encountering emergencies, the big data collaborative scheduling platform immediately activates the intelligent control plan. On the one hand, it quickly notifies nearby energy storage sites to increase their energy storage capacity and store excess electric energy; on the other hand, according to the real-time grid load conditions, it automatically distributes the excess power to surrounding areas with lower loads and transfers it to power consumption nodes in need through flexible DC transmission technology to ensure the efficiency and balance of power supply.

[0041] Preferably, in S4, the following operation steps are specifically included:

[0042] S401: Build a blockchain demand response platform open to the whole society. Based on the distributed ledger characteristics of the blockchain, permanently record all user behavior data participating in demand response, including the time, electricity quantity, and response frequency of reducing electricity consumption, in encrypted form;

[0043] S402: Design an incentive points system. The points obtained by users participating in demand response can not only be exchanged into cryptocurrency at the real-time exchange rate for electricity bill deduction, but also be used to exchange for energy-saving commodities;

[0044] S403: Collaborate with government departments and environmental protection organizations to give additional honorary commendations to corporate users who actively participate in demand response and achieve remarkable results. At the same time, these enterprises enjoy priority green channels in aspects such as environmental protection project approval and green credit application, so as to encourage more enterprises to actively engage in demand response and improve the participation rate.

[0045] A power grid load management system based on big data, adopting the above-mentioned power grid load management method based on big data, includes a cross-industry data access module, a social media public opinion monitoring module, an adversarial network load prediction module, a knowledge graph construction module, and a load management strategy execution module. The cross-industry data access module is used to access the data of financial institutions and real estate, and adjust the power grid electricity load based on the accessed data; the social media public opinion monitoring module is constructed based on an open-source web crawler, which captures the data of social platforms in real time, screens the public opinion information related to electricity consumption, and increases the current regional electricity load based on the public opinion information; the adversarial network load prediction module is based on the constructed model to predict the load trend of the power grid in real time; the knowledge graph construction module is used to construct a knowledge graph and formulate personalized service strategies for users according to characteristics; the load management strategy execution module is used to execute the strategies for power grid power generation planning and allocation according to the content predicted by the adversarial network load prediction module and the regional environment.

[0046] Preferably, the cross-industry data access module establishes dedicated line connections with financial institutions and real estate departments, regularly obtains enterprise credit and capital flow data, removes invalid values and duplicate values through data cleaning algorithms, classifies enterprises by industry, and uses a linear regression model to estimate the change in the electricity consumption demand of enterprises in the next 3-6 months based on the trends of credit growth and capital inflow and outflow; interfaces with the real estate department to receive the delivery progress and occupancy rate estimation information of newly built buildings in real time, and estimates the initial and subsequent growth loads of residential electricity consumption according to the house type structure, supporting facilities, and combined with empirical coefficients using a multiple linear regression algorithm.

[0047] Preferably, the social media public opinion monitoring module uses an open-source web crawler framework, sets city names and activity keywords, captures the data of social media platforms and local life forums in real time, uses the text classification algorithm in natural language processing to screen out the public opinion information related to electricity consumption, and once discovers the information of large-scale event holding, combines the event scale and historical electricity consumption data of similar events, and estimates the newly added electricity load around the event through simple linear regression.

[0048] Preferably, the load management strategy execution module executes the strategy for the grid load based on the data content of the cross-industry data access module, the social media public opinion monitoring module, the adversarial network load prediction module, and the knowledge graph construction module.

[0049] Compared with the prior art, the present invention provides a method and system for grid load management based on big data, having the following beneficial effects:

[0050] 1. The method and system for grid load management based on big data integrate the interactive data of social media, the sales data of ticketing platforms, and the user participation data of local life service platforms, avoiding the one-sidedness of single-dimensional evaluation, being able to more accurately reflect the real popularity of activities, being able to obtain and analyze the data of each platform in real time, being able to timely reflect the change trend of activity popularity, making the evaluation results more timely, facilitating activity organizers and grid staff to adjust strategies in a timely manner, and considering the number of ticket refunds in the calculation of ticket popularity scores, which helps to more accurately evaluate the actual attractiveness and popularity of activities. By setting weights for the data of different platforms and different behavior indicators, it can be flexibly adjusted according to the nature of the activity and the characteristics of the target audience, making the popularity evaluation more in line with the actual situation and improving the accuracy and pertinence of the evaluation.

[0051] 2. The method and system for grid load management based on big data mine the association between social media public opinion and electricity consumption, monitor the dynamics of social media platforms and local life forums in real time, quickly capture the sudden changes in electricity load caused by urban activities and people's livelihood hotspots, can dispatch mobile energy storage vehicles in advance to ensure power supply, and issue peak-shaving power consumption prompts to surrounding residents, realizing the dynamic balance of the grid load. Through intelligent regulation plans, the energy storage can be adjusted, and at the same time, the power can be distributed according to the actual grid load situation, effectively ensuring power supply.

[0052] 3. The method and system for grid load management based on big data use a generative adversarial network to assist load prediction, combine the integrated data, can accurately capture the impact of various complex factors on the grid load, and at the same time, through cross-industry data collaborative collection, can know in advance the changes in electricity demand brought by enterprises and communities. Based on this, the construction of infrastructure such as substations and transmission lines can be planned in advance, reducing the high costs and electricity risks brought by temporary capacity expansion. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 is a flowchart of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0054] In order to make the technical means, creative features, achieved purposes and effects of the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

[0055] Example 1:

[0056] As Figure 1 shown, a power grid load management method based on big data includes the following operating steps:

[0057] S1: Multi-data collection and in-depth integration. Collect and integrate cross-industry data and social media data, and plan the power grid based on the collected data to improve the working load of the power grid;

[0058] Cross-industry data includes cooperating with financial institutions, actively communicating and sharing data, regularly obtaining the credit limit and capital flow of enterprises from financial institutions. Based on the obtained data, through data analysts, according to industry experience and past data models, judge the dynamic production scale of enterprises. When it is seen that the credit limit of a certain manufacturing enterprise has increased by 20% in the past three months and the capital flow is towards the construction of a new production line, mark the enterprise. The power grid planning department starts to investigate the situation of surrounding substations and estimate the expansion demand; Connect with the real estate management department to clarify the information acquisition channels for newly built residential communities and commercial buildings. Starting from six months before the delivery of the building, update the delivery time every month, and estimate the occupancy rate based on the housing type ratio and surrounding supporting facilities. After obtaining these original data, power engineers calculate the initial and subsequent growth loads of residential electricity and public area electricity according to the power standards and usage duration estimation models of different electrical equipment. When the production scale expands, it can provide accurate guidance for the layout optimization of the power grid;

[0059] Social media data includes using natural language processing technology tools to access the public data interfaces of social media platforms and local life forums, setting keywords, including city names and activities, home appliance brands and promotions, and real-time capturing relevant discussion content. Once it is monitored that a large-scale activity is to be held in a certain area, combine the historical electricity consumption data of similar activities and the activity heat evaluation algorithm to calculate the newly added electricity load around the activity, and based on the calculated electricity load, give an electricity consumption reminder for the current area, immediately notify the power grid dispatching department to arrange mobile energy storage vehicles, and at the same time send peak-shaving electricity consumption reminders to surrounding residents through text messages and APP push. The activity heat evaluation algorithm includes the calculation of the social media influence score S, the calculation of the ticket heat score T, and the calculation of the local life service heat score L. Among them, the calculation formula for the social media influence score S is:

[0060]

[0061] where α is the social media weight coefficient, set according to experience and historical data regression analysis, indicating the relative importance of social media in the overall heat evaluation; Like is the number of likes; Comment is the number of comments; Retweet is the number of retweets; w Like is the weight of likes; w comment is the weight of comments; w Retweetis the weight for forwarding; i is the number of posts about the event on social media;

[0062] The formula for calculating the ticket popularity score T is:

[0063] T = β(TicketSold × w TicketSold - TicketRefund × w TicketRefund ) + γTicketSpeed × w TicketSpeed ;

[0064] where β is the ticket sales volume weight coefficient; TicketSold is the ticket sales volume; TicketRefund is the ticket refund volume; w TicketSold is the weight of the ticket sales volume; w TicketRefund is the weight of the ticket refund volume; γ is the sales speed weight coefficient; TicketSpeed is the sales speed; w TicketSpeed is the weight of the sales speed;

[0065] The formula for calculating the local life service popularity score L is:

[0066] L = δ

[0067] (Collect × w collect + Appointment × w Appointment + Browse × w Browse ) ;

[0068] where δ is the local life service platform weight coefficient; Collect is the collection; Appointment is the appointment; Browse is the browse; w collect is the weight of the collection; w Appointment is the weight of the appointment; w Browse is the weight of the traffic;

[0069] The total event popularity H is calculated based on the social media influence score S, the ticket popularity score T, and the local life service popularity score L, and the formula is:

[0070] H = S + T + L;

[0071] The higher the calculated H value, the greater the event popularity. Furthermore, the possible additional electricity load around the event can be predicted based on the popularity, and the grid dispatching preparation can be made in advance;

[0072] Continuously monitor the hotspots in the electricity consumption fields such as new energy vehicles and energy-saving household appliances, analyze the changes in the public discussion popularity. If a certain electricity-consuming device is popular, quickly cooperate with the manufacturer to implant the intelligent electricity consumption mode program in the sales link of the electricity-consuming device to ensure that after purchase, the electricity-consuming device can automatically pre-cool according to the low-load period of the power grid, and at the same time match a preferential electricity price package for users to guide users to use electricity during off-peak hours.

[0073] S2: Data analysis and intelligent modeling. Collect historical electricity consumption data, and construct models and knowledge graphs based on the collected data to predict future power grid electricity load conditions, and at the same time optimize the operation and service strategies of the power grid;

[0074] The load prediction assisted by the generative adversarial network includes the following operation steps:

[0075] S201: Data preparation. Collect historical electricity consumption data of the power grid, covering electricity load information under different seasons, weekdays and holidays, and various weather conditions. At the same time, sort out the factors affecting the electricity load, accurately label the above data according to the time series, and divide the collected data into a training set, a validation set and a test set according to the ratio of 70%, 15%, and 15% to ensure uniform data distribution and comprehensively reflect various electricity consumption scenarios;

[0076] S202: Model construction. Build a generative adversarial network architecture, including a generator and a discriminator. The generator part is constructed based on a deep learning neural network and is used to try to generate data simulation samples similar to real electricity consumption scenarios. The samples cover the key features of the electricity load magnitude and change trend. The discriminator is constructed based on a neural network. The input is the simulation samples generated by the generator and the real historical electricity consumption data samples. By extracting and comparing the features, it outputs a discrimination probability to indicate the possibility that the input sample is a real data sample or a simulation data sample;

[0077] S203: Training and optimization. Let the generator and the discriminator perform adversarial training. During the training process, use the validation set to regularly evaluate the model performance and monitor the similarity index between the simulation data samples and the real data samples. When the performance of the model on the validation set no longer improves, stop the training, use the test set to conduct a final evaluation of the model, compare the predicted electricity load of the model with the actual electricity load, calculate the accuracy rate and recall rate. When the expected effect is achieved, deploy the model to the actual power grid load prediction system to enable the power grid to perform load prediction under abnormal conditions;

[0078] The construction of a user electricity consumption portrait driven by a knowledge graph includes the following operation steps:

[0079] S204: Knowledge graph data collection. Collect basic electricity consumption information of users from the internal systems of power companies, including user IDs, addresses, monthly electricity consumption, and payment records. At the same time, obtain detailed electricity consumption data recorded by users' electric meters, including daily peak and valley periods of electricity consumption and power factors. Collaborate with public security departments to obtain users' ages and family member information, with real estate departments to obtain property areas and housing types, and with financial institutions to understand users' credit status to ensure the legal compliance of data sources. Collect information on the electricity-consuming equipment owned by users. On the one hand, through the registration records when users purchase electrical appliances, and on the other hand, use smart meters to identify the characteristics of newly added electricity-consuming equipment;

[0080] S205: Knowledge graph construction. Determine the entities and relationships of the knowledge graph. Entities include users, electricity-consuming equipment, electricity-consuming areas, and policies and regulations. Relationships include users owning electricity-consuming equipment, users being located in a certain electricity-consuming area, and users applying a certain policy. Based on graph database technology, store and organize the collected data according to the determined entities and relationships to form an interconnected knowledge graph;

[0081] S206: Generation of user electricity consumption portraits. According to the knowledge graph, extract key features of users, including age groups, credit ratings, and the use of new energy equipment. Classify users according to different feature combinations. For different categories of users, formulate personalized electricity consumption service strategies. The strategies include, for elderly user groups with high credit and multiple new energy equipment at home, increasing the peak-valley electricity price discounts and guiding them to use equipment with flexible electricity consumption time adjustment during the low valley electricity price period at night;

[0082] For young entrepreneurial enterprises, design flexible electricity settlement methods based on the industry prospects and financial status associated with the knowledge graph, including allowing enterprises to settle electricity bills quarterly and providing a certain amount of electricity bill deferral period, and then adjusting to a more conventional settlement mode later;

[0083] Provide industrial enterprises with an intelligent equipment management solution. Automatically adjust the start-stop sequence and operation time of large equipment according to the production process and grid load conditions. For example, in a steel production workshop, when the grid load approaches the warning value, the system preferentially suspends auxiliary equipment in non-critical links, such as reducing the frequency of operation of dust removal fans, and preferentially ensures the electricity consumption of core equipment such as steelmaking furnaces; after the load eases, then orderly resume the operation of auxiliary equipment to ensure production continuity while ensuring grid stability.

[0084] S3: Cooperative operation of distributed energy virtual power plants. Establish an energy database and a data collaborative dispatching platform, integrate external data and the energy database in real time, and then adjust the power generation plan of energy in advance and conduct energy storage and dispatching of grid power according to the actual situation;

[0085] Specifically, it includes the following operation steps:

[0086] S301: Conduct a comprehensive inventory of all distributed energy resources within the power grid area, and detailedly record the installed capacity, power generation efficiency, equipment health status, historical power generation data, and grid connection nodes of each energy site to establish a refined distributed energy database;

[0087] S302: Develop a specialized big data collaborative scheduling platform. Using the Kalman filtering algorithm, integrate real-time external data such as weather forecasts, hydrological monitoring, and biomass raw material supply with the information in the energy database. Taking distributed photovoltaic power generation as an example, by accessing high-precision meteorological satellite data, accurately predict the movement trajectory of clouds and changes in light intensity, update the light expectation for the next 2 hours every 10 minutes precisely, and then adjust the power generation plans of other complementary energy sources in advance to ensure stable overall power supply;

[0088] S303: When encountering emergencies, the big data collaborative scheduling platform immediately activates the intelligent control plan. On the one hand, quickly notify nearby energy storage sites to increase their energy storage capacity and store excess electrical energy; on the other hand, according to the real-time load situation of the power grid, automatically allocate excess power to surrounding areas with lower loads and, through flexible DC transmission technology, cross-regionally allocate it to power consumption nodes in need to ensure the efficiency and balance of power supply.

[0089] S4: Establish a blockchain-based demand response incentive mechanism. Utilize the distributed ledger feature of the blockchain to record all behavior data for participating in demand response, and design an incentive points system to encourage users to actively reduce electricity consumption during peak hours;

[0090] Specifically, it includes the following operation steps:

[0091] S401: Build a blockchain demand response platform open to the whole society. Based on the distributed ledger feature of the blockchain, permanently record all behavior data of users participating in demand response, including the time, electricity quantity, and response frequency of reducing electricity consumption, in an encrypted form;

[0092] S402: Design an incentive points system. The points obtained by users participating in demand response can not only be exchanged for encrypted digital currencies at the real-time exchange rate for electricity bill deduction but also be used to exchange for energy-saving commodities;

[0093] S403: Collaborate with government departments and environmental protection organizations to give additional honorary awards to enterprise users who actively participate in demand response and achieve remarkable results. At the same time, these enterprises enjoy priority green channels in aspects such as environmental protection project approval and green credit application to encourage more enterprises to actively engage in demand response and increase the participation rate.

[0094] Example 3:

[0095] A power grid load management system based on big data, including a cross-industry data access module, a social media public opinion monitoring module, an adversarial network load prediction module, a knowledge graph construction module, and a load management strategy execution module. The cross-industry data access module is used to access data from financial institutions and real estate, and adjust the power grid electricity load based on the accessed data; the social media public opinion monitoring module is built based on an open-source web crawler, which can capture data on social platforms in real time, screen out public opinion information related to electricity consumption, and increase the current regional electricity load based on the public opinion information; the adversarial network load prediction module is based on the constructed model to predict the load trend of the power grid in real time; the knowledge graph construction module is used to construct a knowledge graph and formulate personalized service strategies for users according to their characteristics; the load management strategy execution module is used to execute the strategies for power grid power generation plans and allocations according to the content predicted by the adversarial network load prediction module and the regional environment.

[0096] The cross-industry data access module establishes dedicated line connections with financial institutions and real estate departments, regularly obtains enterprise credit and capital flow data, removes invalid values and duplicate values through data cleaning algorithms, classifies enterprises by industry, and uses a linear regression model to estimate the changes in the electricity consumption demand of enterprises in the next 3-6 months based on the trends of credit growth and capital inflows and outflows; it interfaces with the real estate department to receive real-time information on the delivery progress and occupancy rate estimates of newly built buildings, and uses a multiple linear regression algorithm to estimate the initial and subsequent growth loads of residential electricity consumption according to the house type structure and supporting facilities, combined with experience coefficients. For example, for an ordinary two-bedroom apartment, the basic electricity consumption is estimated at 50-80 watts per square meter, plus the shared electricity consumption of elevators, public lighting, etc., and the overall electricity load estimate of the community is calculated comprehensively, and it is dynamically adjusted as the occupancy rate increases.

[0097] The social media public opinion monitoring module uses an open-source web crawler framework, sets city names and activity keywords, captures data on social media platforms and local life forums in real time, and uses text classification algorithms in natural language processing to screen out public opinion information related to electricity consumption. Once it discovers information about large-scale event hosting, it combines the event scale and historical electricity consumption data of similar events, and estimates the newly added electricity load around the event through simple linear regression.

[0098] The load management strategy execution module executes strategies for the power grid load based on the data content of the cross-industry data access module, the social media public opinion monitoring module, the adversarial network load prediction module, and the knowledge graph construction module.

[0099] The foregoing has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and what is described in the above embodiments and the specification is only to illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements fall within the scope of the present invention claimed. The scope of protection claimed for the present invention is defined by the appended claims and their equivalents.

Claims

1. A power grid load management method based on big data, characterized in that: It includes the following operation steps: S1: Multi-data collection and in-depth integration. Collect and integrate cross-industry data and social media data, and plan the power grid based on the collected data to improve the working load of the power grid; S2: Data analysis and intelligent modeling. Collect historical electricity consumption data, and build models and knowledge graphs based on the collected data to predict the future power grid electricity load conditions, and at the same time optimize the operation and service strategies of the power grid; S3: Cooperative operation of distributed energy virtual power plants. Establish an energy database and a data collaborative scheduling platform, integrate external data and the energy database in real time, and then adjust the power generation plan of the energy in advance, and store and dispatch the power grid electricity according to the actual situation; S4: Establish a blockchain-based demand response incentive mechanism. Utilize the distributed ledger feature of the blockchain to record all the behavior data used to participate in the demand response, and design an incentive points system to encourage users to actively reduce electricity consumption during peak hours.

2. The method for power grid load management based on big data according to claim 1, wherein: In S1, the cross-industry data includes cooperating with financial institutions, actively communicating and sharing data, regularly obtaining the credit limit and capital flow of enterprises from financial institutions, and based on the obtained data, through data analysts, according to industry experience and past data models, judging the dynamic production scale of enterprises. When the production scale expands, it can provide accurate guidance for optimizing the layout of the power grid; The social media data includes using natural language processing technology tools to access the public data interfaces of social media platforms and local life forums, setting keywords, including city names and activities, home appliance brands and promotions, and real-time capturing relevant discussion content. Once it is monitored that a large-scale event is to be held in a certain area, the additional electricity load around the event is calculated by combining the historical electricity consumption data of similar events and the event heat evaluation algorithm, and based on the calculated electricity load, a power consumption reminder is given to the current area, and the power grid dispatching department is immediately notified to arrange a mobile energy storage vehicle. At the same time, a peak-shaving power consumption reminder is sent to the surrounding residents through text messages and APP push. The event heat evaluation algorithm includes the calculation of the social media influence score S, the calculation of the ticket heat score T, and the calculation of the local life service heat score L. The calculation formula for the social media influence score S is: Among them, α is the social media weight coefficient, which is set according to empirical and historical data regression analysis, representing the relative importance of social media in the overall popularity evaluation; Like is the number of likes; Comment is the number of comments; Retweet is the number of retweets; w Like is the weight of likes; w comment is the weight of comments; w Retweet is the weight of retweets; i is the number of posts about the event on social media; The calculation formula for the ticket heat score f is: T = β(TicketSold × w TicketSold - TicketRefund × w TicketRefund ) + γTicketSpeed × w TicketSpeed ; Among them, β is the ticket sales volume weight coefficient; TicketSold is the ticket sales volume; TicketRefund is the ticket refund volume; w TicketSold is the weight of the ticket sales volume; w TicketRefund is the weight of the ticket refund volume; γ is the sales speed weight coefficient; TicketSpeed is the sales speed; w TicketSpeed is the weight of the sales speed; The calculation formula for the local life service heat score L is: L = δ (Collect×w collect +Appointment×w Appointment +Browse×w Browse ); where δ is the weight coefficient of the local life service platform; Collect is for collection; Appointment is for reservation; Browse is for browsing; w collect is the weight of collection; w Appointment is the weight of the reservation; w Browse is the weight of the flow rate; Calculate the total event heat H based on the social media influence score S, the ticket heat score T, and the local life service heat score L. The formula is: H = S + T + L; The higher the calculated H value, the greater the event heat. Furthermore, the possible additional electricity load around the event can be predicted based on the heat, and the power grid dispatching preparation can be made in advance; Continuously pay attention to the hotspots in the electricity consumption fields such as new energy vehicles and energy-saving home appliances, analyze the changes in the public discussion heat, and if it is found that a certain electricity-consuming device is popular, quickly cooperate with the manufacturer to implant an intelligent electricity consumption mode program in the sales link of the electricity-consuming device to ensure that after purchase, the electricity-consuming device can automatically pre-cool according to the low-load period of the power grid, and at the same time match a preferential electricity price package for users to guide users to use electricity during off-peak hours.

3. A method for power grid load management based on big data according to claim 1, characterized in that: In S2, it specifically includes generating adversarial network-assisted load forecasting and knowledge graph-driven construction of user electricity consumption portraits. The generating adversarial network-assisted load forecasting includes the following operation steps: S201: Data preparation. Collect historical electricity consumption data of the power grid, covering electricity load information under different seasons, weekdays and holidays, and various weather conditions. At the same time, sort out the factors affecting electricity load, accurately label the above data according to the time series, and divide the collected data into a training set, a validation set, and a test set at a ratio of 70%, 15%, and 15% to ensure uniform data distribution and comprehensively reflect various electricity consumption scenarios; S202: Model construction. Build a generating adversarial network architecture, including a generator and a discriminator. The generator part is constructed based on a deep learning neural network and is used to try to generate data simulation samples similar to real electricity consumption scenarios. The samples cover the key features of electricity load magnitude and change trend; The discriminator is constructed based on a neural network. The input is the simulation samples generated by the generator and the real historical electricity consumption data samples. By extracting and comparing features, it outputs a discrimination probability to represent the possibility that the input sample is a real data sample or a simulation data sample; S203: Training and optimization. Let the generator and the discriminator perform adversarial training. During the training process, use the validation set to regularly evaluate the model performance and monitor the similarity index between the simulation data samples and the real data samples. When the performance of the model on the validation set no longer improves, stop the training. Use the test set to finally evaluate the model, compare the predicted electricity load of the model with the actual electricity load, and calculate the accuracy rate and recall rate. When the expected effect is achieved, deploy the model to the actual power grid load forecasting system to enable the power grid to perform load forecasting under abnormal conditions; The knowledge graph-driven construction of user electricity consumption portraits includes the following operation steps: S204: Knowledge graph data collection. Collect basic electricity consumption information of users from the internal system of the power company, including user ID, address, monthly electricity consumption, and payment records. At the same time, obtain detailed electricity consumption data recorded by the user's electric meter, including daily peak and valley electricity consumption periods and power factor. At the same time, cooperate with the public security department to obtain user age and family member information, obtain property area and housing type structure from the real estate department, and cooperate with financial institutions to understand the user's credit status to ensure the legal compliance of data sources. Collect information on the electricity-consuming equipment owned by users. On the one hand, through the registration records when users purchase electrical appliances, and on the other hand, use smart meters to identify the characteristics of newly added electricity-consuming equipment; S205: Knowledge graph construction. Determine the entities and relationships of the knowledge graph. The entities include users, electricity-consuming equipment, electricity-consuming areas, and policies and regulations. The relationships include users owning electricity-consuming equipment, users being located in a certain electricity-consuming area, and users applying a certain policy. Based on graph database technology, store and organize the collected data according to the determined entities and relationships to form an interconnected knowledge graph; S206: Generation of user electricity consumption portraits. According to the knowledge graph, extract the key features of users, including age groups, credit ratings, and the usage of new energy equipment. Classify users according to different feature combinations, and formulate personalized electricity consumption service strategies for different categories of users. The strategies include increasing the peak-valley electricity price discounts for elderly users with high credit ratings and multiple new energy devices at home, and guiding them to use devices with flexible electricity consumption time adjustment during the low valley electricity price period at night; For young startup enterprises, design flexible electricity settlement methods based on the industry prospects and financial status associated with the knowledge graph, including allowing enterprises to settle electricity bills quarterly and providing a certain amount of electricity bill deferral period, and then adjusting to a more conventional settlement mode later; Provide an intelligent equipment management solution for industrial enterprises, and automatically adjust the start-stop sequence and operation time of large equipment according to the production process and grid load conditions.

4. A method for power grid load management based on big data according to claim 1, characterized in that: In S3, the specific operation steps are as follows: S301: Conduct a comprehensive inventory of all distributed energy resources in the power grid area, and record in detail the installed capacity, power generation efficiency, equipment health status, historical power generation data, and connection nodes to the power grid of each energy site, and establish a detailed distributed energy database; S302: Develop a special big data collaborative scheduling platform, and use the Kalman filter algorithm to integrate external data such as weather forecasts, hydrological monitoring, and biomass raw material supply with the information in the energy database in real time; S303: When encountering emergencies, the big data collaborative scheduling platform immediately activates the intelligent regulation plan. On the one hand, it quickly notifies nearby energy storage sites to increase their energy storage capacity and store excess electric energy; on the other hand, according to the real-time grid load conditions, it automatically distributes the excess power to surrounding areas with lower loads and transfers it to power consumption nodes in need through flexible DC transmission technology to ensure the efficiency and balance of power supply.

5. A method for power grid load management based on big data according to claim 1, characterized in that: In S4, the specific operation steps are as follows: S401: Build a blockchain demand response platform open to the whole society. Based on the distributed ledger characteristics of the blockchain, permanently record all user behavior data participating in the demand response, including the time, electricity quantity, and response frequency of reducing electricity consumption, in an encrypted form; S402: Design an incentive points system. The points obtained by users participating in the demand response can not only be exchanged for encrypted digital currencies at the real-time exchange rate for electricity bill deduction, but also be used to exchange for energy-saving commodities; S403: Cooperate with government departments and environmental protection organizations to give additional honorary awards to enterprise users who actively participate in the demand response and achieve remarkable results. At the same time, these enterprises enjoy priority green channels in aspects such as environmental protection project approval and green credit application, so as to encourage more enterprises to actively participate in the demand response and improve the participation rate.

6. A power grid load management system based on big data, which adopts the power grid load management method based on big data described in the above claims 1-5, includes a cross-industry data access module, a social media public opinion monitoring module, an adversarial network load prediction module, a knowledge graph construction module, and a load management strategy execution module, and is characterized in that: The cross-industry data access module is used to access data from financial institutions and real estate, and adjust the power grid electricity load based on the accessed data; the social media public opinion monitoring module is built based on an open-source web crawler, which captures data on social platforms in real time, screens out public opinion information related to electricity consumption, and increases the current regional electricity load based on the public opinion information; the adversarial network load prediction module is based on the constructed model to predict the load trend of the power grid in real time; the knowledge graph construction module is used to construct a knowledge graph and formulate personalized service strategies for users according to characteristics; the load management strategy execution module is used to execute the strategies for power grid power generation planning and allocation according to the content predicted by the adversarial network load prediction module and the regional environment.

7. A power grid load management system based on big data according to claim 6, characterized in that: The cross-industry data access module establishes dedicated connections with financial institutions and real estate departments, regularly obtains enterprise credit and capital flow data, removes invalid values and duplicate values through data cleaning algorithms, classifies enterprises by industry, and uses a linear regression model to estimate the changes in the electricity consumption demand of enterprises in the next 3-6 months based on the trends of credit growth and capital inflows and outflows; it interfaces with the real estate department to receive information on the delivery progress and occupancy rate estimates of newly built buildings in real time, and uses a multiple linear regression algorithm to estimate the initial and subsequent growth loads of residential electricity consumption according to the housing type structure and supporting facilities, combined with an empirical coefficient.

8. The power grid load management system based on big data according to claim 6, wherein: The social media public opinion monitoring module uses an open-source web crawler framework, sets the city name and activity keywords, captures data on social media platforms and local life forums in real time, and uses text classification algorithms in natural language processing to screen out public opinion information related to electricity consumption. Once information on the holding of a large-scale event is found, combined with the scale of the event and the electricity consumption data of historical similar events, the additional electricity load around the event is estimated through simple linear regression.

9. A power grid load management system based on big data according to claim 6, characterized in that: The load management strategy execution module executes strategies for the power grid load based on the data content of the cross-industry data access module, the social media public opinion monitoring module, the adversarial network load prediction module, and the knowledge graph construction module.

Images